container-scan-dockle

Warn

Audited by Gen Agent Trust Hub on Feb 14, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill instructs the user to install dockle from an external, non-whitelisted repository (goodwithtech/r/dockle). This requires the user to trust a third-party maintainer outside of the predefined trusted organizations list.- [COMMAND_EXECUTION] (LOW): The skill executes shell commands using the dockle binary. While necessary for its function, this grants the agent the ability to interact with the host system's shell.- [INDIRECT_PROMPT_INJECTION] (LOW): The skill audits container images which are external, untrusted sources of data. An attacker could craft an image with malicious metadata (e.g., specific labels or file names) that, when parsed by dockle and subsequently processed by the AI agent, could attempt to influence the agent's behavior.
  • Ingestion points: dockle-results.json (output from scanning an external image).
  • Boundary markers: Absent; the agent is instructed to read and present the JSON results directly.
  • Capability inventory: Shell execution (dockle).
  • Sanitization: Absent; the skill does not specify any sanitization or validation of the scan results before processing.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 14, 2026, 03:23 PM